PMSM의 극 저속 Sensorless 제어를 위한 SMO 설계

Abstract

The PMSM(Permanent Magnet Synchronous Motor) has advantages such as high efficiency, high power factor and power density because there is no excitation winding for generating magnetic flux. Therefore, it receives high attention in control of high-performance. In the speed control of a PMSM, informations such as the rotor speed and angular position are essential. Usually, a physical sensors such as resolvers and encoders are used to obtain the rotor speed and angular position. However, these sensors have problems of cost rise, volume increase and reliability decrease. In particular, the reliability of the sensor is degraded, when the distance between the control device and the PMSM is too long or the control system is located in a harsh environment. Hence, sensorless control technique without physical sensors of the PMSM is required. This sensorless control has the advantage of saving production cost, decreasing volume, and improving reliability. In this paper, the SMO(Sliding Mode Observer) of model based method is applied among various sensorless control algorithms. The SMO is easy to design by using dynamic model of PMSM, and is characterized by robustness against variation of system parameters, disturbance and noise. Also, the SMO estimates back-EMF(Electromotive force) from the current and voltage information and calculates the rotor speed and angular position through back-EMF. The SMO has good performance in the middle and high speed area, but sensorless control is too difficult in the extremely low speed area. Because chattering problem at back-EMF occurs due to the signal to noise ratio reduction of the current information. Therefore, it is very important to suppress the chattering problem in back-EMF in order to realize sensorless control in very low speed area. Design factors of the SMO affecting chattering problem are observer gain, Low Pass Filter(LPF), phase compensator, and PI gains of speed and current controller. Previous researcher applied saturation function, adaptive observer gain, cascade LPF with variable cut-off frequency and phase compensator to suppress chattering problems at back-EMF. Their comprehensive control performance has been confirmed, but their individual analysis is not done. Especially, the MATLAB simulation was done only in the continuous system, so discrete system characteristic analysis is insufficient. In this paper, the control performance of SMO design factors is analyzed through MATLAB simulation and experiment in a discrete system. Finally an optimized SMO design factors are proposed for extremely low speed sensorless realization.